| Bridge-type grab ship unloader is the main large-scale mechanical equipment for port bulk carrier handling.Its safe and stable operation plays a vital role in the operation and production of enterprises.With the continuous improvement of working time and strength of port bridge-type grab ship unloader,the failure of ship unloader is correspondingly increased.At present,the existing fault diagnosis technology of ship unloader can effectively reduce dangerous accidents,but it is still unable to find fault problems and solve them in time.In view of the shortcomings of the existing fault diagnosis research,such as the untimely and incomplete fault diagnosis,this paper obtains the monitoring data by monitoring the operation status of the bridge grab ship unloader with sensors,and realizes the fault prediction of the ship unloader by using the new fault prediction method.Starting from the characteristics and changes of mechanical state of bridge grab ship unloader,this paper studies the mining of association rules and the prediction of deep confidence network,which makes a useful exploration for the fault prediction of the whole machine system and the trolley operation system of ship unloader.The main work of this paper is as follows.(1)The mechanical structure and working characteristics of the bridge grab ship unloader are analyzed.The main fault types and fault characteristics of the ship unloader are summarized.Based on the above,two online unloader monitoring programs are designed.Through the online monitoring,a large number of ship unloader condition monitoring data has been accumulated,which provides a data source for fault prediction,and also ensures that all the research results in this paper are true and reliable.(2)Aiming at the failure of the bridge grab ship unloader,a ship unloader fault prediction model method based on interest degree association rules is proposed.The sensor monitoring and time domain analysis method are used to obtain the operating parameter space of the ship unloader.The clustering discrete algorithm is used to discretize the monitoring data into nonlinear clustering intervals according to its attribute range,obtain the ship unloader association rule group,and extract the state data association rights.The weight coefficient is constructed,and the state monitoring data association rule directivity feature constraint function model is constructed.The fault prediction is realized by predicting the state change of the association rules in the model.Experiments show that the method can effectively characterize the associated internal feature information of the ship unloader’s operating state monitoring,and realize the prediction of the ship unloader fault category,which has practical significance for reducing the frequency of the ship unloader fault.(3)Based on the fault of the most critical car operating system of the bridge grab ship unloader,a time-series fault prediction model based on Deep Blief Net(DBN)was built.Combining the self-learning ability of deep learning theory,the original time domain signal data is input into DBN for training,and the whole fine-tuning is carried out through reverse fine-tuning learning.The model is used to predict the vibration intensity time series of ship unloader and multi-step prediction to construct condition monitoring.The residual sequence feature constraint function model realizes fault prediction by correlating the state changes of residual weight residual sequences.By comparing and analyzing other related fault prediction methods,the experiment verifies that the method can reduce the early warning threshold and improve the prediction accuracy,and can predict the failure type of the ship unloader early and effectively,and make warnings in advance.(4)Based on the bridge grab ship unloader machinery,sensors,fault prediction technology,etc.,a new type of reliable bridge grab unloader working status monitoring and early warning system was developed.The system has good reliability,real-time and application value,and realizes the early warning functions such as condition monitoring,data storage,fault prediction,fault evaluation,early warning release and fault handling of the ship unloader. |
